On Wed, Nov 12, 2014 at 12:59 PM, Luca Barbato <lu_z...@gentoo.org> wrote:
> On 12/11/14 11:06, Vittorio Giovara wrote:
>> On Tue, Nov 11, 2014 at 6:08 PM, Luca Barbato <lu_z...@gentoo.org> wrote:
>>> On 11/11/14 17:40, Vittorio Giovara wrote:
>>>> CC: libav-sta...@libav.org
>>>> Bug-Id: CID 700465 / CID 1240206
>>>> ---
>>>> libavcodec/lzwenc.c | 5 ++++-
>>>> 1 file changed, 4 insertions(+), 1 deletion(-)
>>>>
>>>> diff --git a/libavcodec/lzwenc.c b/libavcodec/lzwenc.c
>>>> index 7c37bf2..2533f49 100644
>>>> --- a/libavcodec/lzwenc.c
>>>> +++ b/libavcodec/lzwenc.c
>>>> @@ -129,7 +129,7 @@ static inline int findCode(LZWEncodeState * s, uint8_t
>>>> c, int hash_prefix)
>>>> int h = hash(FFMAX(hash_prefix, 0), c);
>>>> int hash_offset = hashOffset(h);
>>>>
>>>> - while (s->tab[h].hash_prefix != LZW_PREFIX_FREE) {
>>>> + while (h >= 0 && s->tab[h].hash_prefix != LZW_PREFIX_FREE) {
>>>> if ((s->tab[h].suffix == c)
>>>> && (s->tab[h].hash_prefix == hash_prefix))
>>>> return h;
>>>
>>> h can be negative?
>>
>> Rereading the code there is an assert() in findCode() which initializes h.
>> Should that be replaced with an averror?
>>
>
> Yes please if it is not purely theoretical
I believe having to carry the averror may cumbersome in a hash function.
The assert will prevent and BadUse anyway, so I'll mark this as false positive.
--
Vittorio
_______________________________________________
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
